Vivian Berger Mediator

Home « Mediation, Employment, Civil Rights « When Mediation Fails

What Happens After Mediation “Fails”?
A Personal Account

NYSBA Labor and Employment Law Journal
June 2021


I have been a mediator for more than 35 years. Like many in my day, before mediation became as popular and widespread as now, I received my first training at a community mediation center. The cases I handled, as a volunteer, were not commercial and did not require a law degree; ordinarily, the parties came unrepresented. I oversaw matters as varied as landlord-tenant disputes, fallings out among neighbors (sometimes involving both parents and children), custody and “P.I.N.S.”[1] proceedings referred from Family Court, and lovers’ quarrels. Because I worked in Queens County, the archetypical city melting-pot, the clients’ origins were as diverse as the cases themselves:  Black, White, Latinx, Asian, straight, gay, Orthodox Jewish. I then branched out into mediating complaints against police officers that had been filed with the New York City Civilian Complaint Review Board. I also made my first foray into the world of employment mediation; for a short time, I took U.S. Postal service cases and a few matters pending before the New York State Division of Human Rights. In these early years, during most of which I was still an active Columbia Law School professor as well an occasional litigator, I viewed mediation as an avocation, not a career, and was content with my miscellaneous and principally unpaid or low-paid docket.

Fast forward to 2000: I retired from Columbia Law School that year, having taken two week-long general trainings and some shorter ones in specific subject matter areas. Around then, I began to focus on developing a full-time ADR practice and decided to become a specialist in employment matters. In the criminal law domain, to which I had devoted the bulk of my academic and litigation efforts until that point, there was always a story — none of it was boring. I figured, correctly, that the same would be true when at least one party was an individual, not a company. Again, working for various court panels and the Equal Employment Opportunity Commission (“E.E.O.C.”), for quite a long time I mainly volunteered my services or got paid below-market rates.

Now, two decades later, a seasoned neutral, I remain on the mediation panel of the United States Court for the Eastern District of New York. The bulk of my work is private, however, or comes to me via the American Arbitration Association (“A.A.A.”).

The Current Project:  Whatever Became of Those Cases I Could Not Resolve?

I have always been personally curious what happens after a mediation of mine “fails” — my shorthand for not settling either at the mediation or soon thereafter, in its wake.[2] And enough of my professor identity survives for me to be also intellectually curious. As many of you know, post-Columbia I continue to write; most of my articles relate to Alternative Dispute Resolution or employment or their intersection.[3] All have been informed by my practice. A number have also relied in large part on empirical data,[4] with PACER providing an especially rich resource for federal court litigation.

For this piece, I decided to use my own records of employment mediations, from approximately 2000 to date, as an actual database rather than just a source of general insights or particular anecdotes. I retrieved 435 cases, virtually all in the areas of discrimination or wage and hour. In addition to contemporaneous notes, which always showed if a matter had settled during a session or shortly afterward, I often had longer-term post-mediation data in the file. From time to time over the years, I had tried to learn the “end of the story” — how and when the action or dispute had come to an end, through negotiation, dismissal or, very infrequently, trial. Now, in much more organized fashion, I supplemented my earlier findings by consulting a number of external sources, such as PACER, the New York State Supreme Court web site, other internet resources, and counsel.[5] My aim was to locate those cases that had reached a definitive  conclusion other than settlement: for example, involuntary dismissals with prejudice, either by summary judgment or otherwise, a hearing on the merits (as occurs at the E.E.O.C. in some federal sector proceedings), a trial verdict, or an arbitration award.

I acknowledge up front the limitations of this study. Among other things, my earlier records are less detailed that the later ones. (I had to cull my papers in order to avoid turning my home office into a welter of filing cabinets.) Moreover, in matters disposed of in forums other than the local federal courts — state courts, arbitrations or the E.E.O.C. — or ones that never were litigated, tracing post-mediation history at times presented steep obstacles. Such problems caused some sample bias. Further, the most recent mediations, as well as a few that proceeded at a glacial pace, posed a “tail” or “censoring problem: “In outcome studies of cases filed within a study window, the sample may be skewed if many cases are still pending when the window closes and not included in the study.”[6] Last, different types of employment disputes, like discrimination and wages and hours, might not be fungible,[7] thereby making the results less susceptible to general conclusions. But even with these caveats,[8] I will next analyze what I found to see if the data were at least suggestive of any (probably modest) conclusions.

Results of the Study

Of the 435 cases with which I began, those with results known to me that did not settle during mediation or in its immediate aftermath, and were not involuntarily dismissed with prejudice, almost all settled in the end. Most likely, this was also true of the relatively few that I could not trace.[9] I was able to track 29 (6.7%) of the total to a final, non-settlement disposition. (Fortunately, all involved discrimination, not wages and hours, claims — thereby avoiding the apples-and-oranges comparison problems discussed in Part II.)

My principal finding was that in 25 (86.2% of 29), the dispute ended adversely to the plaintiff. Nineteen grants of summary judgment made up 65.5% of the 29; two sui generis dismissals with prejudice swelled the ratio of such dismissals to 72.4%. The remaining four instances of claimant loss (13.8% of 29) occurred either after a federal court trial (three cases) or after an E.E.O.C. hearing (one). Finally, the employee prevailed a scant four times out of 29 tries (again, 13.8%), three times in federal court trials and once in arbitration.

Evaluating My Findings

The more extreme the results of even a small sample, the likelier they are to be reliable. This insight can be applied to the 86.6% overall rate of defeat, which befell claimants after a mediation with me failed to produce a resolution; the same should hold true for its converse, the merely 13.8% rate of success.  What do I draw from these results?

My Overall Conclusion (and a Non-Conclusion)

First, let me be clear what I do not conclude: I do not argue that the best course is always for the plaintiff to settle at mediation, if at all possible. For one thing, the employee may succeed in getting a better offer later: for example, after additional discovery,[10] doing well at her deposition, or surviving summary judgment.  Because of confidentiality provisions in the vast majority of discrimination complaint settlements, the mediator usually cannot learn the terms of any subsequent deal.[11] (It is debatable whether the recent changes in New York State law, barring a defendant from imposing such a constraint on an unwilling complainant,[12] has substantially altered the practice: some claimants may themselves want confidentiality, and others may negotiate a premium for agreeing to it.)

Moreover, judgments about the wisdom of an employee’s decision to settle implicitly assume that the plaintiff is, or should be, an “economic man”: a person who measures costs against benefits and acts in accord with his rational self-interest. But individuals may have non-economic reasons to “go for broke” rather than resolve their cases. For instance, they may gain a sense of empowerment by standing up to the defendant.  They may seek even an improbable victory in order to teach the employer a lesson and, it is hoped, prevent others from suffering discrimination in the future. Or they may wish to inflict pain on their adversary by forcing it to spend a lot of money in mounting a defense.[13] At least, if a claimant understands the risks she confronts and is conscious of the basis for her decision — rather than, say, so blinded by emotion that she is not really making a knowing choice among options — she is acting in her own interests, whether or not these appear rational in strictly economic terms.

Furthermore (and this is why, in discussing resolution at mediation, I inserted the qualifier “if at all possible”), settlement may not be a realistic choice for the plaintiff if the defense takes a no-pay or nuisance value position in a case where that stance essentially amounts to bad faith or is wholly unreasonable in light of the probable objective value of the matter.  Note, however, that for defendants as well as plaintiffs, non-economic motives may ground a refusal to settle on sensible terms. Some employer decision makers, like some claimants, are guided by emotions like anger or hurt and, at least initially, insist: “Millions for Defense, Not One Cent for Tribute.”[14] Or they may be acting on financial considerations that extend beyond the complaint at hand: for example, the wish to gain a reputation for never settling in the view that this will save them money in the long run by deterring future lawsuits.

Having explained my non-conclusion, that the plaintiff’s best option is always to resolve the dispute at mediation if she can, I now turn to my actual conclusion: if, for whatever reason, she decides to prosecute her case till adjudication by a judge or a jury, she is extremely likely to lose. Of course, not settling at mediation, even if given a “ballpark” offer, does not necessarily indicate a firm resolution never to settle, as I have discussed. Yet mediation often affords the best opportunity for a complainant to achieve closure on acceptable terms. After all, a disinterested professional is helping the parties. And, if the mediation is early, the employer, or its liability insurer, will not have expended a lot of money on litigation — which might otherwise, more fruitfully, have contributed toward a settlement.

My Reaction to My Findings

Naturally, I was unsurprised to learn that complainants claiming employment discrimination face daunting obstacles; indeed, I have often said that expressly, and it is hardly a novel insight.[15] What did shock me somewhat, at first, was the height of the mountain a plaintiff would have to scale in order to achieve victory.

But, on reflection, I should not have been astonished since my own previous research shows that, on average, plaintiffs have a roughly 38.9% chance of surviving a summary judgment motion (at least, in our local federal courts)[16] and a 30.0% chance of obtaining a favorable verdict at trial[17]. In other words, the combined overall rate of success is a mere 11.7% (.389 times .330), even less than the 13.8% probability revealed in this account of my failed mediations. In addition, focusing just on summary judgment, the 65.5% rate of loss experienced by my claimants — which can be inversely expressed as a 34.5% survival rate — comes close to the 38.9% rate documented in my earlier study of the subject.[18]

Are There Any Lessons Here?

A key lesson for claimants and, thus, for mediators, too, is that a realistic assessment of early cases must take explicit and serious account of the odds of defeating summary judgment, not just those of prevailing at trial[19]: taking a percentage of a percentage rapidly yields a very small chance of ultimate victory. Although one might think this conclusion obvious, certainly to experienced counsel, the mediator ordinarily needs to work hard to focus lawyers, not to speak of clients, on the negatives of their cases. These include not only any problems specific to the matter at hand but also ones presented by the legal system in general: as pertinent here, an arguably too great predilection to cut employees off at the pass, denying them an opportunity to present their proof to the finder of fact. (Anecdotally, this tendency may be less marked in state court than federal.) I recommend that, for any complaint already filed and assigned to a judge, counsel try to ascertain the jurist’s record in deciding summary judgment motions in employment disputes. As preparation for my mediations, I frequently do so myself in an effort to supplement my overall emphasis on the risk of pre-trial dismissal.

However, I stress that I do not criticize lawyers as a group for resisting reality-testing — at least, initially. It is only human nature to view one’s own cause with rose-colored glasses, and attorneys are human too. Such bias is a major reason mediators are needed so often.[20]

Why Did My Failed Cases Not Settle in Mediation?

Finally, I looked at my case notes to see if I could detect any pattern to the causes of failure in my mediations. Although each matter presented its own peculiar dynamics, more often than not actions taken by the claimant’s side scotched resolution. When impasse occurred, it was frequently because either the employee started “too high” — at times, so much so that the employer declined to respond, thereby depriving the plaintiff of the chance to elicit a potentially acceptable offer — or, in the end, held out for “too much.” I use the quoted normative terms not just because 20-20 hindsight revealed that the game was not worth the candle, but also because my evaluation of the case at the time indicated that I believed the plaintiff was taking a very big gamble and making a mistake (at least, if she hoped to maximize her monetary gain).

Sometimes, the lawyer overvalued the case[21] ; often the client did, disregarding the mediator’s cautions and, I suspect, her attorney’s advice. Occasionally, both client and counsel had drunk the same Koolaid, or the latter may have feared losing the former’s confidence by pushing too hard. In no small number of instances, however, defendants created the barrier to settlement by taking a no-pay position or offering only nuisance value. While later events (often in the form of summary judgment) may have validated their view of the complaint’s absence of merit, usually any consideration of their prospective attorneys’ fees would have justified offering more.

A Note on Plaintiffs’ Few Victories on the Merits

In the three federal trial victories for the complainant, he or she initially got a blow-out verdict: millions of dollars. Even following post-trial motions and appeals, and exclusive of attorneys’ fees and costs, the plaintiffs ended up, respectively, with approximately $500,000, $900,000 and just shy of $2,750,000. In the latter two cases, one involving a very appealing disabled claimant and the other, a victim of repeated and gross sexual harassment, it was evident to me from the outset that the defendants, notwithstanding seasoned counsel, had seriously underestimated the risks they faced. (Yet, in addition, as I contemporaneously noted, referring to the harassed woman: “I think maybe this is a case that must be tried because plaintiff feels she has nothing to lose and wants to roll the dice and ‘get back at’” her oppressor.) Indeed, in the matter of the young man with a disability, the defense did not deign to bargain at all. By contrast, the race discrimination claimant who eventually won half a million dollars may just have gotten lucky. He had made a demand so high — though, in retrospect, not irrational — and his attorney behaved in such a toxic manner at the mediation that the employer’s offer of only low nuisance value was understandable, if misconceived.


Having excavated the data yielded by my practice, I hope that this exploration might stimulate other neutrals to do the same. Because of the need for confidentiality, we mediators cannot discuss our cases anecdotally in any detail, and especially not in print. Analyses of aggregate, anonymous data, such as I have conducted before,[22] may generate broader-based insights helpful to other practitioners as well as scholars, but they lack the specificity of a link to a single source. The present piece amounts to an effort to bridge the gap between the particular and the general: one professional’s mining of the raw material of her mediations of employment conflicts. The exercise has given me considerable personal satisfaction. If it has also fulfilled my aim of refining this ore into a modest tool for others, so much the better.


  1. “P.I.N.S.” stands for Persons In Need Of Supervision.  These are young people over whom their parents have lost control; truancy is the most common complaint. ...back

  2. Even in such instances, mediation can, to be sure, have many beneficial effects; some are emotional, others pragmatic.  For example, the reality testing that occurred in this setting may have left its mark on the parties and counsel, influencing later, more successful, bargaining.  Thus, the term “failure” is a bit overbroad, although convenient. ...back

  3. See "Selected Articles on Mediation, Employment, and Civil Rights". ...back

  4. See, e.g.,Vivian Berger, Winners and Losers: Employment Discrimination Trials in the Southern and Eastern Districts of New York: 2016 Update, 42 NYSBA Labor & Emp. L.J. 39 (Spring 2017) [hereinafter “Trials Update”]; Vivian Berger, Punitive Damages in Employment Discrimination Cases: Myth or Reality?, 37 NYSBA Labor & Emp. L.J. 6 (Fall/Winter 2012); Vivian Berger, Michael O. Finkelstein & Kenneth Cheung, Summary Judgment Benchmarks for Settling Employment Discrimination Lawsuits," 23 Hofstra L. & Emp. L.J. 45 (2005)....back

  5. In keeping with my usual custom of focusing on matters arising in Second Circuit states (the better to educate local practitioners), I excluded from my “universe” a few cases from other jurisdictions like New Jersey. ...back

  6. Michael O. Finkelstein, Bruce Levin, Ian M. McKeague, and Wei-Yann Tsai, A Note on the Censoring Problem in Empirical Case-Outcome Studies, 3 J. Empirical Legal Stud. 375, 375 (2006).  And on account of changes in New York City law during this period that favor the discrimination plaintiff (a term I use interchangeably with “claimant” or “complainant”), a relative deficit of later cases could introduce additional problems of sample bias.  Given the extreme paucity of trials, these would likelier affect summary judgment outcomes.  Happily, the matters that I know to be still outstanding amount to only 2.8% of the total. ...back

  7. Discrimination cases can look very different from each other as well.  For instance, non-monetary relief is often a more significant feature of requests for relief by federal sector employees, who remain on the job while litigating much more often than other plaintiffs. ...back

  8. Cf. Stephen B. Burbank, Drifting Toward Bethlehem or Gomorrah? Vanishing Trials and Summary Judgment in Federal Civil Cases, 1 J. Empirical Legal Stud. 591, 593 (2004) (noting temptation to substitute unreliable data for no data). ...back

  9. Altogether, including the matters that I resolved, I am aware of 368 settlements; they constitute 84.6% of the initial database.  Since we have a culture of settlement, this outcome is unsurprising. ...back

  10. That being said, I think that plaintiffs tend to overestimate the chance of turning up the proverbial smoking gun in discovery.  They also, at times, place too much value on what they perceive as smoking guns, which they already have by the mediation.  Increasingly, these come in the form of contemporaneous tape recordings.  Even when audible, they are not always as devastating as employees and their lawyers believe. ...back

  11. The Second Circuit’s requirement of court approval of FLSA settlements, see Cheeks v. Freeport Pancake House, Inc.,796 F.3d 199 (2d Cir. 2015), means that, by contrast, in most wage and hour cases there will be a public record of the parties’ agreement. ...back

  12. GOL § 5‑336 (effective, as amended, Oct. 11, 2019). ...back

  13. Typically, the claimant will have a contingent fee arrangement with her attorney. ...back

  14. See Vivian Berger, They’re Human, Too: The Care and Feeding of Defendants in Employment Mediations, 39 NYSBA Labor & Emp. L.J. 5, 6 (Winter 2014). ...back

  15. See generally Vivian Berger, Book Review, 43 NYSBA Labor & Emp. L. J. 18 (Spring 2018) (reviewing Ellen Berrey, Robert L. Nelson & Laura Beth Nielsen, Rights on Trial: How Workplace Discrimination Law Perpetuates Inequality (2017)); see also id. at 18 (“It is no secret to anyone knowledgeable in the field that employment plaintiffs have a very difficult roe to how ....”) (citation omitted). ...back

  16. See Berger, Finkelstein & Chung, supra note 4, at 60. ...back

  17. “Trials Update,” supra note 4, at 40. ...back

  18. See Berger, Finkelstein & Chung, supra note 4. More granular analysis reveals that 22 of the 29 cases in our data base were litigated in federal court. These accounted for 16 (72.7%) of 22 summary judgment dismissals or, conversely, 27.3% of surviving federal cases — a figure less close to, and lower than, the statistic derived from my case database: 38.9%. On the one hand, the former, 27.3%, is more on point since my summary judgment study also used a federal court data base. On the other hand, the latter, 38.9%, comprised represented cases only, and one of my former employee clients was pro se at the time of the motion. That status would have lowered his chance of avoiding dismissal. See id. at 55 (Table 1). Furthermore, the smaller the universe of cases considered (22 versus 29, the denominator in our ratio), the less likely that any conclusions drawn from it can be generalized. This caveat applies in spades to my finding here that four out of eight complainants eventually won on the merits — six after trial in federal court and two at an arbitration hearing — a 50% win rate, in contrast to the 30.0% figure in my earlier study. See “Trials Update,” supra note 4, at 40. Moreover, the latter did not examine arbitration awards. ...back

  19. Naturally, my underlying assumption is that the employer will make a summary judgment motion. While I lack statistics on the actual prevalence of such motions, experience teaches that they are extremely common in employment cases when the matter has not been dismissed or otherwise resolved by the close of discovery. ...back

  20. I sometimes tell mediation participants that I am not necessarily the smartest person in the room — nor am I, surely, the most knowledgeable about their conflict. But, happily, I do not need to be, because I am surely the most neutral. ...back

  21. A few plaintiffs’ firms have a policy of starting in the stratosphere. Usually, though, the attorney will come down to earth if the opening demand does not cause the defense to simply refuse to counter. ...back

  22. See supra text at note 4 and note 4. ...back